Inbound Preight Consolidation: A Simulation Model to Evaluate
Consolidation Rules
Daniel J. Ford Jr.
Bachelor of Arts, Economics
Davidson College, 2001
Submitted to the Engineering Systems Division in
Partial Fulfillment of the Requirements for the Degree of
MASTER OF ENGINEERING IN LOGISTICS
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
JUNE 2006
ARCHIVES
O Daniel Ford All rights reserved.
The author hereby grants to MIT permission to reproduce and to distribute publicly paper and
electronic copies of this thesis document in whole or in part.
Signature of Author
May 23,2006
Certified by
..................................................
/'
Accepted by .....................................................................
[
....
Yosef Sheffi
Professo ngineeri Systems Division
Professor, Civil and Environmental Engineering Department
Director, MIT Center for Transportation and Logistics
Inbound Preight Consolidation: A Simulation Model to Evaluate
Consolidation Rules
Daniel J. Ford Jr.
Submitted to the Engineering Systems Division
on May 23,2006 in Partial Fulfillment of the Requirements for the
Degree of Master of Engineering in Logistics
Abstract
In logistics, freight can be consolidated over time (temporally) or over space (spatially). This
thesis presents a simulation model to evaluate temporal and spatial consolidation rules. The
model is the result of a research project to analyze freight consolidation options for a large
industrial company. The research project focused on the company's fieight imported from China
to the US, and the model presented in the thesis is structured to represent a typical import
logistics network.
The results section of the thesis presents a method for evaluating consolidation rules. The results
recommend temporal consolidation of two weeks at the origin port and temporal consolidation of
less than one week at the factory for the company's shipments from China to the US. This
consolidation policy offers total network cost savings of 24% over the base case, an immediate
ship policy.
Thesis Supervisor: Edgar Blanco
Title: Research Associate
Table of Contents
...........................................................................................................6
1. INTRODUCTION
...........................................................................................................................................
1.1. Benefits of Consolidation
7
1.1 . 1 Returns to Scale ....................................................................................................................................................7
1.1.2 Reduction of Fixed Costs ..................................................................................................................................... 7
1.1.3 Speed and Reliability........................................................................................................................................... 8
...............................................................................................................................................
1.2 Costs of Consolidation
8
1.2.1 Time ...................................................................................................................................................................... 8
1.2.2 Distance................................................................................................................................................................8
1.2.3 Coordination Costs ............................................................................................................................................... 9
2 . INDUSTRY CASE
.......................................................................................................10
..............................................................................................................................................10
2.2 Data Summary ..........................................................................................................................................................11
2.1 Operational Overview
3. LITERATURE REVIEW
.............................................................................................
13
.......................................................................................................13
3.2 Monte Carlo Simulation ..........................................................................................................................................15
3.'1 Results from Previous Consolidation Studies
.
4 SIMULATION MODEL
.................................................................................................16
....................................................................................................................................17
4.2 Temporal Consolidation Rules ...............................................................................................................................22
4.3 Production Output Simulation ...............................................................................................................................23
4.4 Transit Time Simulation..........................................................................................................................................24
4.1 Spatial Consolidation Rules
4.4.1 Factory to Origin Port ......................................................................................................................................... 24
4.4.2 Origin Port to Destination Port ..........................................................................................................................25
4.4.3 Destination Port .................................................................................................................................................. 26
4.4.4 Destination Port to DC .......................................................................................................................................27
................................................................................................................................................................
4.5 Constraints
28
4.5.1 Factory to Origin Port ........................................................................................................................................-28
4.5.2 Origin Port to Destination Port ..........................................................................................................................28
4.5.3 Destination Port to DC .......................................................................................................................................28
..............................................................................................................-30
4.7 Fixed and Variable Transit Costs...........................................................................................................................
31
4.6 Walk Through of the Simulation Model
4.7.1 Factory to Origin Port .........................................................................................................................................31
4.7.2 Origin Port to Destination Port ..........................................................................................................................31
4.7.3 Destination Port: ................................................................................................................................................-32
4.7.4 Destination Port to DC .......................................................................................................................................32
................................................................................................................................................
4.8 Cost of Transit Time
35
4.8.1 Pipeline Inventory Holding Cost ....................................................................................................................... 35
4.8.2 Safety Stock Inventory Holding Cost ................................................................................................................ 35
5. INDUSTRY CASE RESULTS ....................................................................................
37
...............................................................................................................................................................39
5.2 Total Transit Cost and Inventory Cost..................................................................................................................42
5.3 Transit Cost. Transit Time. and Transit Time Variance (Tradeoffs) ...............................................................44
5.1 Transit Cost
6. SUMMARY OF RESULTS ..........................................................................................49
7. FUTURE RESEARCH ...............................................................................................5 0
..................................................................................................................51
REFERENCES
APPENDIX ........................................................................................................................52
Table of Exhibits
Exhibit 2.1. Samples of Historical Shipment Data .......................................................... 11
Exhibit 2.2. Annual Volume (kgs) ..................................................................................... 1 2
Exhibit 3.1 : Guidelines for Logistic Systems Simulation ...................................................15
libit 4.1 : Spatial Consolidation Cases ............................................................................ 17
libit 4.2. Factory Consolidation (FACT) ...................................................................18
libit 4.3 : Origin Port Consolidation (ORIG) ..................................................................19
zibit 4.4. Destination Port Consolidation (DEST) .......................................................... 20
libit 4.5. Origin Port and Destination Port Consolidation (FULL) ................................ 21
libit 4.6. Production Output Simulation (sample) .......................................................... 23
libit 4.7. Port to Port Transit Simulation (sample) .........................................................25
libit 4.8. Customs and Freight Release Simulation (sample) .........................................26
libit 4.9. Inland Truck Time Simulation (sample) .........................................................27
libit 4.10. Simulation Model Flowchart .....................................................................2 9
libit 4.1 1: Cost Drivers and Reference Tables .............................................................. 3 4
libit 4.12. Pipeline and Safety Stock Inventory Drivers..............................................3 6
zibit 5.1: Consolidation Rules for Results Discussion .................................................... 38
libit 5.2. Transit Cost Graphs................................................................................... 39
..
libit 5.3 : Transit Cost Summary.................................................................................... -40
libit 5.4. Transit Cost Savings Summary ....................................................................... 41
libit 5.5. Transit Cost + Inventory Cost Graphs ............................................................. 42
libit 5.6. Transit Cost + Inventory Cost Summary ......................................................... 43
libit 5.7. Transit Cost + Inventory Cost Savings Summary ...........................................43
libit 5.8. Transit Cost vs. Transit Time Graph ............................................................... 44
libit 5.9. Transit Cost vs. Transit Time (Best Options Graph) .......................................46
libit 5.10. Transit Cost vs . Transit Time Variance (Best Options Graph) ......................47
Exhibit A.1:
Exhibit A.2:
Exhibit A.3:
Exhibit A.4:
Exhibit A.5:
Exhibit A .6:
Reference Table 1. Factory to DC (100 rows) ............................................... 52
Reference Table 2. Factory to Origin Port (100 Rows) .................................. 53
Reference Table 3. Origin Port to Destination Port (9x9. 8 1 rows) ................54
Reference Table 4. Destination Port to DC (9 rows) .....................................55
Reference Table 5. Destination Port (9 rows)................................................ 57
Reference Table 6. DC (9 rows) .....................................................................58
1. Introduction
A graduate student mails letters home to his mother everyday. The problem is he simply can't
afford the post. For this reason he has decided to implement a consolidation program to save
money on the stamps, but he is still analyzing what consolidation method will allow him to save
money while still maintaining the letter's timely value.
The graduate student has two options for consolidation: temporally consolidation (over time) or
spatial consolidation (over space). To consolidate the letters temporally, he holds today's letter
for shipment with tomorrow's. By waiting one day, he is able to ship two letters in one envelope
and save the cost of one stamp. To consolidate spatially, he meets his sister (who also mails
letters home to mom daily) at the post office, where they can combine letters to ship in one
envelope, again saving the cost of one stamp.
The purpose of this thesis is to present a model that evaluates rules for spatial and temporal
consolidation. If the graduate student waits to send letters home to mom, how long should he
wait? If he and his sister decide to meet at the post office, which post office should they meet at?
The case of sending letters home to mom is not far removed from the choices that many
company's face today in shipping. This thesis is the result of a sponsored research project to
analyze freight consolidation options for a large industrial company. The simulation model
presented in this thesis is the direct result of the research project.
First, the thesis will introduce the industrial company's current situation. Second, the thesis will
present the simulation model. Finally, the results of the research project will be highlighted and
discussed. But before we get there, let's first review some of the benefits and costs of
consolidation.
1.1 Benefits of Consolidation
1.1.1 Returns to Scale: Freight logistics exhibit significant returns to scale. For instance,
UPS charges about $15 to ship a 1-pound package overnight, but only about $30 to ship a 15pound package overnight. The per unit freight rate for the 1-pound package is about $15/lb,
while the per unit freight rate for the 15-pound shipment is about $2/1b. Although this is an
extreme example, returns to scale exist throughout freight logistics. Whether comparing lessthan-truckload (LTL) to truckload (TL), less-than-container (LCL) to full container (FCL), or
10,000 kg barges to 100,000kg barges; per unit freight rates decrease as the shipment size
increases.
1.1.2 Reduction of Fixed Costs: Consolidation reduces the total # of shipments, which in
turn reduces total fixed costs. Regardless of the situation, companies incur fixed costs
throughout the shipping process. In the letter to mom example, there is a fixed cost every time
the student walks to the post office. In the case of UPS, it takes a fixed amount of time to fill out
a UPS delivery form. In the case of an import network, a customs broker receives fixed
payments independent of the shipment size. Consolidation reduces the number of shipments,
thereby reducing the number of times that you have to walk to the post office, fill out a UPS
form, or pay fixed brokerage charges.
1.1.3 Speed and Reliability: Consolidated shipments can exhibit increased speed and
reliability. For instance, full truckload (FT) is generally faster and more reliable (transportation
time variance is less) than less-than-truckload (LTL) shipments. Likewise, processing at US
ports (the time from arrival until availability) is general faster and more reliable for full container
shipments (FCL) than for less-than-container (LCL) shipments. Moreover, goods shipped by FT
or FCL are touched less and are less prone to damage than goods shipped by LTL or LCL.
1.2 Costs of Consolidation
1.2.1 Time: Despite all of its benefits, consolidate is not without cost. Temporal consolidation
consumes time and cost savings from consolidation must be balanced against the value of time.
The cost of transit time is directly related to the value of the goods in transit and the variability of
the transit time.
In this model, transit time starts when the goods finish production and ends with the goods arrive
at the DC. In contrast, transportation time (as discussed in 1.1.3) starts when the good leave
point A and ends when the goods arrive at point B. Transportation time does not include the
time waited at point A. Thus, although consolidated shipments exhibit increased speed and
reliability in transportation time, this time savings is often lost in during the waiting process.
1.2.2 Distance: Spatial consolidation requires that goods travel a greater distance. In the case
of sending letters home to mom, though the student and his sister both live in Boston, the post
office nearest to his house is different from the post ofice nearest to her house. Spatial
consolidation thus requires that one of them will have to travel farther than if they shipped
individually. Increases in distance must be accounted for.
1.2.3 Coordination Costs: Finally, consolidation requires coordination and rules. If the
student and his sister are going to meet at the post office, they need to coordinate the time and
location to meet. Coordination takes time and effort. Systematic consolidation rules can help
reduce coordination costs. For instance, the student and his sister could set a rule to meet at post
office "A" at 3:00 pm every Wednesday. Nevertheless, analyzing the network to determine
appropriate rules still takes time and effort.
2. Industry Case
The sponsor company for this thesis, a large industrial corporation, produces the majority of its
products in Asia. The majority of these products are produced in China. For this reason, we
decided to focus our research on consolidation options for the company's inbound freight from
China.
2.1 Operational Overview
According to interviews with the company's logistics manager, the company's suppliers and
factories in China ship more of less when they feel it's appropriate. Shipments are sent directly
from the factory to the DC with no option for consolidation the origin port or the destination port.
According to our understanding, informal consolidation is in place at the factory, but
consolidation of any type outside of the factory does not exist.
The company uses various transportation providers to ship goods from of China. The company
plans to move all freight to one 3PL provider by year end. Of the providers that the company is
considering, all have the capability to consolidate freight at the China Port and at the US Port.
The freight being shipped is dense and industrial. Shipments "weigh out'' rather than "volume
out" and for this reason the data used for the research project is based on weight rather than
volume.
Finally, most of company's distribution centers are engaged in light manufacturing (tooling,
painting, kitting, and assembly). For this reason, the research project does not include the final
customer. The DC is considered the final customer in the project.
2.2 Data Summary
The simulation model that we will visit later requires volume, cost and time data for the entire
network. Much of the project was spent compiling this data, but in the end, most of the data still
suffered from lack of normalization and completeness. The volume, costs and time distributions
in the model are made using best estimates from the data provided. Finally, although we found
possible evidence for seasonality in the historical shipping data (see exhibit 2. I), the lack of
complete data made it difficult to determine if low seasons were the result of missing data or low
demand. The company's logistics manager noted that demand for the majority of the business is
stable year round. The production output simulation in this thesis assumes no seasonality.
Exhibit 2.1 : Samples of Historical Shipment Data
Factory 37
- Historical Shipments
9000
8000
8 7000
6000
" 5000
a
E 4000
3 3000
2000
1000
0
China Port: Shanghai
Historical Shipments
i
a
-
g
0
a a ( D
US Port Atlanta:
Historical Shipments
Date
e
&
$
Z
eZ
Y
Np
$
DATE
Date
T q F T q T T T $
o,c33$!Ycumd
90000
80000
70000
60000
50000
40000
30000
20000
10000
0
Distribution Center 5
Historical Shipments
$
'
g??eTTs$$$
7 z 2 Date
As we can see from Exhibit 2.2 below the company has approximately 50 manufacturing
facilities or suppliers located in China (across all divisions), although most of the volume is
dominated by five suppliers. The company has eight distribution centers in the US, although
most of the volume from China is received by two DCs. The company exports primarily through
two China ports (Ningbo and Shanghai) and imports primarily through two US ports (Chicago
and Atlanta). Annual shipment weight from China to the US is approximately 3,000,000 KGS 1
Year.
Exhibit 2.2: Annual Volume (kgs)
ANNUAL QUANTITY BY ORIGIN PORT
1600000
1400000
1200000
1-
$
800000
6
W
4
m
2
m
0
DLC
NGB
SHA
SZX
TAO
ORIGIN PORT
FACTORY
lm
ANNUAL QUANTITY BY DC
ANNUAL QUANTITY BY DESTINATION PORT
2000000
180MX)O
1600000
1400000
1200000
$
'C: 1000000
800000
600000
400000
-- --..,
--
1
-----A
200000
0
200000
0
ATL
-
--
2000000
1800000
1600000
1400000
1200000
1000000
800000
600000
400000
CHI
HOU
LAX
NYC
DC1
DC2
DC3
DC4
DC5
DESTINATION PORT
DC
El
m
DC6
DC7
DC8
DC9
3. Literature Review
Most of the research written on consolidation focuses on finding an optimal solution for a two or
three node, steady state network. A four node network, with variable demand, variable transit
times, and different modes of transportation, as presented in this thesis is more difficult to
optimize. In order to maintain the realism of a complex model, while at the same time keeping
simplicity in computation, we have chosen to use Monte Carlo simulation rather than
optimization.
First, let's review the results from previous consolidation optimization studies. Considering that
we were unable to find recent simulation studies published on consolidation, the results from
optimization studies on consolidation will serve as a baseline to analyze the feasibility of the cost
savings presented in the industry case results section. After we review these results, we will
introduce guidelines for Monte Carlo simulation used in the framework for the model presented
in this thesis.
3,1 Results from Previous Consolidation Studies
Buffa and Munn (1989) present the case of a commodity product shipped over a two node, one
segment network. The study assumes that only one mode of transport is available, and the key
notion of their model is that "shipping costs is based on freight rate, which is a fbnction of
shipping weight, which, in turn, is dependent on the cycle time for the reorder" (p. 370). In
juxtaposition, the simulation model in this thesis assumes two types of transportation for each
segment, but does not include the order cycle as a decision. Buffa and Munn use a recursive
algorithm to solve for the optimal order cycle time, and they find a "9.2% reduction on the cost
for the set where each item was ordered separately, and a 0.4% reduction on the cost for the set
where all items were ordered together" (p. 374).
Centinka and Lee (2002) present an optimization model to minimize transportation and inventory
costs over a two node, one segment network in "the case where the aggregate demand of the
market area is constant and known per period" (p. 53 1) and transportation time is assumed to be
negligible (p. 532). In contrast, the model in this thesis presents the case where demand is
variable and transit time is significant. Centinka and Lee find that for the "uncapacitated
problem the average cost savings in using the optimal policy, rather than the approximate policy
is given by 3.4%. The maximum cost savings is 13.3% whereas the minimum cost savings is
0.27%" (p. 548).
Nass, Dekker, and Sonderen-Huisman (1 997) focus on determining the optimal "cutoff order
size" for "physical distribution management for a product company in Western Europe7'(p.
1057). They considers the case of a three node network (factory, DC, customer) in which the
factory has the option to ship direct to the customer or through a DC. In contrast, the model
present in this thesis assumes that all shipments flow through each of the four nodes (factory,
origin port, destination port, DC). There is no option to skip the origin port or the destination
port. While Nass et al. focus on the optimal cutoff order, the model in this thesis looks to
evaluate the effect of the consolidation cutoff value. In the case study presented by Nass et al.,
they find an optimal solution which recommends "an increase in delivery through a DC from 6375%" and obtains "2% lower costs" (p. 1063).
Cost savings in the above research varies from a minimum of 0.27% to a maximum of 13.3%.
As highlighted above, the networks in the research above are less complex than the network
modeled in this thesis. Given that complexitydrives costs, we expect the potential cost savings
to be greater in the network at hand than in research found above.
3.2 Monte Carlo Simulation
Monte Carlo Simulations have been used for management decisions for well over 50 years
(Malcom, 1960). Recently, new tools such as Genetic Algorithms and Markov Chain Monte
Carlo have taken Monte Carlo to new heights. This thesis provides a showcase that with basic
modern software (Excel), a 50-year-old simulation technique (Monte-Carlo) can still be used to
create simple, customized, and effective simulations for management decisions.
The below guidelines from Geisler and Steger (1963) are an excellent reference for simulation
design. If a simulation model doesn't have the majority of characteristics listed below, it's likely
an inadequate representation of the system.
Exhibit 3.1: Guidelines for Logistic Systems Simulation
S~sternsCharacteristics
'1) They contain many interacting elements.
2) They contain elements affected by randomness, unpredictability, risk, etc.
3) They include activities whose performance is affected by time lags.
4) Logistics systems require resources, ie costs.
5) Logistics systems require policies, rules, and problem-solving capabilities for their operation.
6) Logistics systems employ information and data.
7) Logistics systems embody component organizations.
8) Logistics systems have mutual impacts with other systems without systems, such as combat
commands, factories, and the like.
Source: Geisler (I 963)
4. Simulation Model
This simulation model is designed to analyze consolidation options for inbound freight from
China to the US. Although the model is focused on a specific country to country import network
(China to the United States), it is applicable to any 4-node, 3-segment import network.
The 4 nodes in the model are:
1. Factory (F)
2. Origin Port (OP)
3. Destination Port (DP)
4. Distribution Center (DC)
The 3 segments in the model are:
1. Factory to Origin Port
2. Origin Port to Destination Port
3. Destination Port to Distribution Center
The model contains 5 main elements:
1. Spatial and Temporal Consolidation Rules
2. Production Output and Transit Time Simulations
3. Constraints
4. Fixed and Variable Transit Costs
5. Cost of Transit Time
The key to this model is its use of spatial and temporal consolidation rules. These rules
determine how freight flows through the network. First, we will review spatial and temporal
consolidation rules. Second, we will explore production output and transit time simulations.
Third, we will look at the model's constraints.
After reviewing these first three elements, we will stop to examine a diagram that shows how
these elements interact to make up the physical structure of the model. Finally, we will review
the financial structure of the model: fixed and variable costs and the cost of transit time.
4.1 Spatial Consolidation Rules
Spatial Consolidation Rules determine the route freight follows in the network. In the model
each factory is linked to a primary origin port and a consolidation origin port. Likewise, each
distribution center is linked to a primary destination port and a consolidation destination port.
Depending on spatial consolidation rules factories either ship to their primary origin port or to
their consolidation origin port, while distribution centers either receive from their primary
destination port or from their consolidation origin port.
This thesis will examine 4 spatial consolidation cases: Direct Shipment, Origin Port
Consolidation, Destination Port Consolidation, and Origin and Destination Port Consolidation. .
The consolidation logic for each case is listed in the table below.
Exhibit 4.1 : Spatial Consolidation Cases
Case
Abbreviation
'FACT
ORIG
DEST
FULL
Case Name
Factory Consolidation
Origin Port Consolidation
Destination Port Consolidation
Origin and Destination Port Consolidation
Rule 1:
Origin Port
Consolidation?
No
Yes
No
Yes
Rule 2:
Destination Port
Consolidation?
No
No
Yes
Yes
The above case abbreviations will be used throughout the paper. To get a better feeling for these
four spatial consolidation cases, the networks diagrams for each of the four cases. A description
of each case is listed below the case network diagrams.
Exhibit 4.2: Factory Consolidation (FACT)
Map Source: CIA
Under direct shipment, there is no spatial consolidation. Factories ship by truck to the China port
closest to them. From the China port goods are shipped by ocean to the US port closest to the
distribution center. The goods are picked up at the US port are trucked to the distribution center.
Factory consolidation is analogous to the student and his sister shipping fiom independent post
offices.
The China ports fiom the top down: Dalian (DLC), Qingdao (TAO), Shanghai (SHA), Ningbo
(NGB), Shenzhen (SZX). US Ports: Los Angeles (LAX), Houston (HOU), Chicago (CHI),
Atlanta (ATL), New York (NYC).
-
Exhibit 4.3: Origin Port Consolidation (ORIG)
Fo1
op1
DC1
DP2
-
DC2
DP3
DC3
,
DP,
L
DC,
Map Source: CIA
Under origin port consolidation, factories ship by truck to their assigned consolidation port (in
this case, Shanghai). From Shanghai, the goods are shipped by ocean to the US port closest to
final distribution center. The goods are then picked up at the US port and trucked to the
distribution center. Origin port consolidation is analogous to the graduate student and his sister
meeting at the post office. Please note, as seen in the diagram above, that due to the feasibility of
long distance trucking in China factories in northern China do not consolidate in Shanghai.
Exhibit 4.4: Destination Port Consolidation (DEST)
Map Source: CIA
Under destination port consolidation, factories ship by truck to the closest China port. From the
China port goods are shipped by ocean to the assigned US consolidation port, in this case Los
Angeles. From LA, goods are shipped by long haul truck to the distribution center. Destination
port consolidation is analogous to the student and his sister shipping large packages
independently to their mother, who then distributes the contents of the package to aunts, uncles,
and grandparents who live in the area.
Exhibit 4.5: Origin Port and Destination Port Consolidation (PULL)
Map Source: CIA
Under full consolidation, factories ship by truck to their assigned China consolidation port.
From the China port, goods are shipped by ocean to the assigned US consolidation port. From
the US consolidation port, goods are shipped by long haul truck to the distribution center. Full
consolidation is thus a combination of China port and US port consolidation. Full consolidation
is analogous to the graduate student and his sister meeting at the post office to ship one large
package to their mother, who then distributes the contents of the package to aunts, uncles, and
grandparents who live in the area.
4.2 Temporal Consolidation Rules
Regardless of the spatial consolidation rules set, we always have the option to wait. Temporal
consolidation rules determine:
1. Max Wait Time: How long we are willing to wait?
2. Cutoff Value: What we are waiting for?
For example, we can set temporal consolidation rules so that the factory ships to the origin port
only iE
1. Any factory inventory is older than 7 days, or
2. The factory inventory is greater than 10000kgs.
In this case the factory cutoff value is 10000kgs, while the factory max wait time is 7 days.
The model is designed to evaluate temporal consolidation rules at the Factory, Origin Port, and
Destination Port. The Industry Case Results section we will compare a range temporal
consolidation rules for each of the four spatial consolidation cases diagramed above.
4.3 Production Output Simulation
In the model, production output contains the following values: Factory, Quantity, Date, and
Distribution Center. In our current case historical shipment data is inadequate to estimate
production output, as informal consolidation is already in place at the factory. As the sponsor
company was unable to provide data on order placement or production output, we estimate the
average order size as ?4of the shipment size from historical data. In the production output
simulation, quantity is based on a normal distribution (mean = average order size, standard
deviation = 20% of the average order size). All figures in the model are restricted to positive
numbers. Production output simulation inputs are referenced from the factory to DC table
(Exhibit A.6). A sample of the production output simulation is found in the exhibit below.
Exhibit 4.6: Production Output Simulation (sample)
FACTORY
F0 1
DC
DC9
1-Jan
0
2-Jan
0
3-Jan
0
4-Jan
0
5-Jan
0
6-Jan
0
The above simulation sample can be read as follows:
January 2: 70 kgs finish production at factory F04 and are ready for shipment to DCl .
January 4: 1041 kgs finish production at factory F04 and are ready for shipment to DC8; 13327
kgs finish production at Factory F06 and are ready for shipment to DC5.. .
4.4 Transit Time Simulation
4.4.1 Pactory to Origin Port: The model assumes that factory to origin port shipments take
one day and exhibit zero variance in transit time. Most of the factories used by the sponsor
company are located near the coast of China, within 3-4 hours of their primary port. The
assumption that each factory to origin port shipment takes 1 day and exhibits zero variance is
reasonable for shipments to the primary port. One the other hand, the distance fi-om the factory
to the consolidation port is father away, and the assumption of zero variance on transit time to
the consolidation port is more tenuous. In an effort to keep the size of the model manageable we
have left out the transit time simulation for the factory to origin port transit, and the assumption
made to avoid this extra simulation should be noted.
4.4.2 Origin Port to Destination Port: Based on limited data from the sponsor company,
FCL and LCL transit time distributions are estimated for each origin port to destination port lane.
Time is measure from origin port departure to destination port arrival and does not include
customs clearance or availability processing. FCL and LCL transit time distributions are
referenced from the origin port to destination port table (Exhibit A.3). A sample of the
production output simulation is found in the exhibit below. Notice that simulation values only
return if a shipment has been made. If no shipment is made, the simulation returns a transit time
of zero.
Exhibit 4.7: Port to Port Transit Simulation (sample)
LANE
OP1
DPI
OP1
DP2
OP 1
DP3
OP1
DP4
OP1
DP5
OP2
DPl
OP2
DP2
OP2
DP3
OP2
DP4
OP2
DP5
OP3
DP1
OP3
DP2
OP3
DP3
OP3
DP4
OP3
DP5
7-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
29
0
8-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
9-Deb
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
10-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
0
25
I 1-Feb
0
0
0
0
0
23
0
0
0
0
0
0
0
0
0
12-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
The above simulation sample can be read as follows:
February 7: A shipment is made from origin port 3 (OP3) to destination port 4 (DP4). This
shipment takes 29 days.
February 10: A shipment is made from origin port 3 (OP3) to destination port 4 (DP5). This
shipment takes 25 days.
4.4.3 Destination Port: FCL and LCL customs clearance and availability processing time
distributions are estimated for each destination port. Time is measured from destination port
arrival to freight availability. The estimated customs clearance and availability processing
distributions are referenced from the destination port table (Exhibit A.5). A sample of the
customs clearance and availability processing simulation is found below.
Exhibit 4.8: Customs and Preight Release Simulation (sample)
LANE
OP 1
DP 1
OP 1
DP2
OP 1
DP3
OP 1
DP4
OPI
DP5
OP2
DPl
OP2
DP2
OP2
DP3
OP2
DP4
OP2
DP5
OP3
DPl
OP3
DP2
OP3
DP3
OP3
DP4
OP3
DP5
DPl
OP4
OP4
DP2
OP4
DP3
OP4
DP4
7-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
4
0
0
0
0
0
8-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
9-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
10-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
0
5
0
0
0
0
11-Feb
0
0
0
0
0
3
0
0
0
0
0
0
0
0
0
0
0
0
0
12-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
13-Feb
0
0
0
0
0
0
0
0
0
0
0
0
0
0
4
0
0
0
0
The above simulation sample can be read as follows:
February 7: A shipment is made OP3 to DP4. This shipment takes arrives at DP4 (after allotted
port to port transportation time) and takes 4 days for customs clearance and freight release.
February 10: A shipment is made OP3 to DP5. This shipment takes arrives at DP5 (after
allotted port to port transportation time) and takes 5 days for customs clearance and freight
release.
4.4.4 Destination Port to DC: FT and LTL transit time distributions are estimated for each
destination port to DC lane. Time is measured from availability at destination port to arrival at
DC. Estimated distributions are referenced from the destination port to DC table (Exhibit A.6).
A sample of the destination port to DC truck time simulation is found below.
Exhibit 4.9: Inland Truck Time Simulation (sample)
Lane
DP1
DP1
DP 1
DP2
DP3
DP3
DP3
DP4
DP5
DC3
DC4
DC2
DC5
DC6
DC7
DC8
DCl
DC9
19-Feb
0
0
0
0
0
0
0
0
2
20-Feb
0
1
0
0
0
0
0
0
0
21-Feb
0
0
0
0
0
0
0
0
0
22-Feb
0
0
0
3
0
0
0
0
1
23-Feb
0
1
0
0
0
0
0
0
0
24-Feb
0
0
0
0
0
0
0
0
0
The above simulation sample can be read as follows:
February 19: A shipment is available for pick up at destination port 5 (DP5) for shipment to
DC9. The truck shipment from DP5 to DC9 takes 2 days.
February 20: A shipment is available for pick up at destination port 1 (DP1) for shipment to
DC4. The truck shipment from DP1 to DC4 takes 1 days.
4.5 Constraints
4.5.1 Factory to Origin Port: Full truckload shipments are constrained by a maximum
truckload weight (1 7000kgs). The model does not include truckload volume as a constraint, as
the case at hand involves industrial freight which weighs out. The model assumes zero
constraints on factor to origin port truck availability.
4.5.2 Origin Port to Destination Port: Ocean shipments are constrained by ocean vessel
availability. Ocean vessel availability is based on a weekly schedule and is referenced from the
origin port to destination port table (Exhibit A.3). The model assumes that factories have access
to ocean vessel shipping schedules and thus only ship to the origin port if a vessel is available.
Full container shipments are constrained by a maximum container weight (17000kgs). The
model does not include container volume as a constraint, as the case at hand involves industrial
freight which weighs out. Finally the model assumes that all freight is shipped by 20' containers.
4.5.3 Destination Port to DC: Destination full truckload shipments are constrained by
maximum truckload weight. Considering that the case involves dense industrial freight which
weighs out, to simplify the model, we do not include a maximum truckload volume constraint.
The model also assumes zero constraints on destination port to DC truck availability.
Exhibit 4.10: Simulation Model Flowchart
ORIGIN PORT
CONSOLIDATION?
DESTINATION PORT
CONSOLIDATION?
LINK
ORlGIN FT
CONSTRAR\TT
IS FnQPn INV >
FT CUTOFF
IS FnOPn INV >
MAX WAIT TIME
FACTORY(Fn)
INVENTORY
(Fn.QPn:DPxDCn)
VESSEL
SCHEDULE
DATE-
PRODUCTION
SIMULATION
FACTORY(Fn)
\ PRODUCTION OUTPUT C n k
(FnQPnDPnDCn)
FnQPn TRUCK
SHIPMENTS
IORIGIN
7
PORT?
IS OPrxDPn IN V >
FCLCUTOFF
FCL
CONSTMINT
ORIGIN PORT(0Pn)
INVENTORY
(OPnDPn-DCn)
@
OPnDPn FCL
SHIPMENTS
-
IS DPxDCn INV >
@
IS OPnDPn INV >
MAXWAITTIME
OPxDPn LCL
TRANSIT TlME
SIMULATION
FCL PORT TO PORT
TRANSIT TlME
SIMULATION
FCL CUSTOMS AND
FREIGHT RELEASE TIME
SIMULATION
-
FREIGHT
FREIGHT
DPn CUSTOMS. FREIGHT
EL CUSTOMS, FREIGHT
FCL
DESTINATION PORT (DPn)
INV AVAILABLE FOR PICKUP
(DPn:DCn)
DESTINATION FT
CoWTITWNT
DPn to DCn LTL
SHIPMENTS
INLAND FT TRANSIT
TIME SIMULATION
I
DISTRIBUTION CENTER
1
I
IS DPrxDCn INV >
/
@
INLAND LTL
TRANSIT TIME
SIMULATION
3.6 Walk Through of the Simulation Model
The above diagram shows how goods flow from the factory to the distribution center. Let's walk
through each of the step in the model:
1)
2)
3)
41
5)
6)
7)
8)
1'
10)
I 1)
12)
13)
14)
15)
10)
17)
18)
19)
20)
2 1)
22)
Production output simulation releases quantities for each factory to DC lane by day.
Spatial consolidation rules set the factory to origin port and distribution center to destination
port links. With the links set, the production output simulation now contains the
quantity, date, factory, origin port, destination port, and distribution center for each
output quantity.
Production output is stored in factory inventory according its origin port to destination port
to distribution center lane.
Factory checks if inventory on hand is greater than full truck cutoff value.
Factory checks if inventory on hand is older than max wait time.
if 4 or 5 are true, factory checks if vessel is available at origin port.
If 6 is true, factory ships to the origin port and goods are stored at the origin port according
to their destination port to distribution center lane.
3PL checks if goods at origin port (for shipment to a specific destination port) are greater
than the full container cutoff value.
3PL checks to see if goods at origin port (for shipment to a specific destination port) are
older than max wait time.
If 8 is true, 3PL ships up to the full container constraint.
If 8 is true, transit time simulation assigns a transit time for FCL shipment.
If 9 is true, transit time simulation assigns a transit time for LCL shipment.
After allotted transit time, FCL and LCL shipments arrive at destination port.
If FCL shipment has arrived, customs and freight release time simulation assigns time to
availability for FCL shipment.
If LCL shipment has arrived, customs and freight release time simulation assigns time to
availability for LCL shipment.
After allotted customs and freight release time goods are available for pickup.
3PL checks if goods at destination port for shipment to a particular DC are greater than the
fir11 truck cutoff value.
3PL checks if goods at destination port for shipment to a particular DC are older than the
max wait time.
If 17 is true, 3PL ships up to full truck constraint.
if 17 is true, Destination Port to DC time simulation assigns full truck transit time.
If 18 is true, Destination Port to DC time simulation assigns less than truck transit time.
After allotted time for inland truck transit, LTL and FT shipments arrive at DC.
As we have seen from the walk through above, rules and simulations govern the flow of goods
through the simulated network. In the model, simulations are based on historical time and output
distribution. These distributions are considered fixed. Changing transportation time
distributions or order output distributions are not decision variables in the model. The only
decision variables are thus temporal and spatial consolidation rules.
The model's fixed and variable costs are based on cost drivers from the simulated network. The
simulated network is based on the fixed simulation distributions and the decision variables, the
temporal and spatial consolidation rules. The model's costs are divided into 1) fixed and
variable transit costs and 2) transit time costs.
4.7 Fixed and Variable Transit Costs
The fixed and variable costs for each segment in the model (or node where applicable) are listed
below.
4.7.1 Factory to Origin Port: For factory to origin port shipments, full truck is assumed to
be the only method of transportation. The cost driver is the number of factory to origin port
truck shipments. Origin truck prices are referenced from the factory to origin port table (Exhibit
A.2). Although some factories are Ex-works and some are Ex-Country, the model includes
truck costs for all factory to origin port shipments.
4.7.2 Origin Port to Destination Port: For origin port to destination port shipments,
transportation options are restricted to full container and less-than-container ocean shipments.
The FCL cost driver is the number of origin port to destination port FCL shipments. The LCL
cost driver is the weight shipped over each origin port to destination port lane by LCL. Port to
Port LCL and FCL prices are referenced Erom the origin port to destination port table (Exhibit
A.3).
Based on interviews with the logistics manager at the industrial company, most 3PL's used by
company will hold freight in China for two weeks free of charge. Based on the situation, we
restricted the max wait time to 14 days or less at the origin port. Accordingly, origin port storage
charges are not included in the model.
4.7.3 Destination Port: Each FCL or LCL shipment, regardless of size is charged a fixed
brokerage charge at the destination port. Based on interviews with the company's logistics
manager, the company pays a flat fee for brokerage regardless of the port. We have set this fee
at $200. The fixed brokerage charge is referenced from the destination port table (Exhibit AS).
4.7.4 Destination Port to DC: For destination port to DC shipments, transportation options
are restricted to FT and LTL shipments. The FT cost driver is the number of destination port to
DC FT shipments. The LTL cost driver is the total weight shipped for each destination port to
DC lane by LTL. LTL costing accounts for volume discounts. FTL and LTL costs are
referenced from the destination port to DC table (Exhibit A.4).
Based on interviews with the company's logistics manager, most trucking companies used by the
company will hold freight in their port warehouse for 1 week free of charge. It should be noted
that cases in the results sections with destination port max wait times greater than 7 days, do not
include the appropriate charges and should be viewed accordingly.
4.7.5 DC: Each FT or LTL shipment, regardless of size, is charged a fixed receiving charge at
the DC. Based on discussions with the company's logistics manager, fixed receiving cost is
estimated at $100 for each DC. Fixed receiving costs are referenced from the DC table (Exhibit
A.6).
Exhibit 4.11 below summarizes the cost drivers and reference tables for all costs described above.
Exhibit 4.11: Cost Drivers and Reference Tables
COST
DRIVER
COST
# OF TRUCK-
OMGIN
SHIPMENTS
-
TRUCKING COSTS
# OF FCL
SHIPMENTS
# OF LCL
REFERENCE
TABLE
FCL OCEAN
FREIHT COSTS
\
PORT TO PORT
---------
SHIPMENTS
LCLOCEAN
FREIHT COSTS
/
-
(FCL +LCL)
FIXED BROKERAGE
COSTS
# OF FTL
SHIPMENTS
FTL INLAND
FREIGHT COSTS
# OF SHIPMENTS
SHIPMENTS
LTL SHIPMENT
# OF SHlPMENTS
(FTL +LTL)
FACTORY TO
ORIGIN PORT
OMGIN PORT
DESTINATION
PORT TO DC
LTL INLAND
FREIGHT COSTS
RECEIVING \
DC
FIXED COSTS
4.8 Cost of Transit Time
Time is of the essence in supply chain management. The cost of transit time includes four main
ingredients. The model accounts for the first two ingredients in the list below:
1. Pipeline Inventory Holding Cost
2. Safety Stock Inventory Holding Cost
3. Lost Sales due to Stockouts
4. Lost Potential for Sales Increase due to Quicker Time to Market
4.8.1 Pipeline Inventory Holding Cost depend on the average transit time, the average value
of goods in transit, the company's holding cost rate, and the quantity shipped per year. The
model uses the below equation to estimate pipeline inventory holding cost:
Pipeline Inv Cost =
where
E[T]*v*r*Q/365
E[T] = average transit time
v = average value of goods / kg
r = company's holding cost rate
Q = quantity shipped per year
4.8.2 Safety Stock Inventory Holding Cost
Safety Stock Inventory Holding Cost is based on a required service level, the variance of demand,
the expected demand, the average lead time of supply, the variance of the lead time of supply,
the value of the goods, and company's holding cost rate. The model uses the below equations to
estimate safety stock inventory holding cost:
Safety Stock Inv Costs = z*ox*r*v
where
z = norminv of (1-company's cycle service level)
v = average value of goods / kg
r = company's holding cost rate
ox = d (E[L]*O;'+E[D]'
where
* 0;')
E[L] = expected lead time of supply
oo2= variance of demand / day
E[D] = expected demand / day
02= variance of lead time of supply
In the model, standard deviation of demand is estimated at 20% of demand. Service Level is
estimated at 95%. Holding cost rate is estimated at 18%. Production lead time is assumed to be
constant at 60 days with zero variance. The expected lead time of supply is equal to 60 days plus
the average transit time. The variance of lead time of supply is equal to variance of transit time
(variance of production lead time is assumed to be zero). The average value of goods for each
DC is estimated at $25 / kg.
A summary of the pipeline and safety stock holding cost drivers
arc listed in Exhibit 4.12 below.
Exhibit 4.12: Pipeline and Safety Stock Inventory Drivers
PIPELINE
INVENTORY COSTS
QmSHIPPED I
AVERAGE
COST OF GOODS
YEAR
AVERAGE LEAD
TIME OF SUPPLY
INVENTORY C O S T S 4
PRODUCTlON
LEAD TlME
VARIANCE OF
VARIANCE OF LEAD
TlME OF SUPPLY
HOLDING COST
RATE
SERVICE LEVEL
TRANSIT TIME
VARlANCE
PRODUCTION
VARIANCE
5. Industry Case Results
The results are divided into three sections:
1) Transit Cost
2) Transit Cost + Inventory Cost
3) Transit Cost, Transit Time, and Transit Time Variance
In the first section, we ignore transit time, transit time variance, and inventory cost to focus
solely on transit cost. This first section explores returns to scale and reduction of fixed costs. As
a reminder, transit cost refers to all fixed and variable cost (exclusive of inventory holding cost)
that the company incurs from the end of production until goods arrive at the DC.
In the second section, we assume that the drivers of inventory cost are known and correct. By
adding inventory cost to transit cost, we arrive at a total network cost to evaluate across the 4
consolidation cases: Direct Shipment, Origin Port Consolidation, Destination Port Consolidation,
and Origin Port and Destination Port Consolidation.
In the third section, we assume that the drivers of inventory cost are unknown or incorrect and
we remove inventory cost fiom the equation. In this third section, we compare (a) the tradeoff
between transit cost and average transit time and (b) the tradeoff between transit time and transit
time variance. As a reminder, transit time refers to the time fiom end of production until goods
arrive at the DC.
All results discussed in this section are based on the consolidation settings listed below in
Exhibit 5.1.
Exhibit 5.1: Consolidation Rules for Results Discussion
CASE ABBREVIATION
PACT
ORIGIN
Origin Port Consolidation?
NO
YES
Destination Port Consolidation?
NO
NO
Factory Cutoff Value
1600Okgs 16000kgs
Factory Max Wait Time
DV1
DV 1
16OOOkgs 16OOOkgs
Origin Port Cutoff Value
DV2
DV2
Origin Port Max Wait Time
Destination Port Cutoff Value
16OOOkgs 16OOOkgs
Destination Port Max Wait Time
1 day
1 day
DEST
NO
YES
160OOkgs
1 day
16OOOkgs
DV1
16OOOkgs
DV2
FULL
YES
YES
16OOOkgs
1 day
16OOOkgs
DV1
16OOOkgs
DV2
As noted in the exhibit above, the cutoff value will be set at 16000kgs for all transportation
modes, while the rnax wait time will serve as the key decision variables (DV) for each
consolidation case. Please note that the factory max wait time is set at 1-day for the destination
port and 1 1 1 consolidation cases. Please note that the destination port rnax wait time is set to 1day for the factory and origin port destination cases.
In this discussion, we will use (DV1,DV2) notation to indicate decision variable settings.
Decision variable 1 (DV1) and decision variable 2 (DV2) for each case are listed above in
Exhibit 5.1. For instance, factory consolidation decision variable settings of (14,8) indicates a
rnax wait time at the factory (DV1) of 14 days and a max wait time at the origin port (DV2) of 8
days.
Throughout the results we use immediate ship policy as the base case. Under immediate ship
policy, once goods finish production they are shipped immediately and do not consolidate at the
origin port or destination port. Under this policy: factory, origin port, and destination port max
wait times are all equal 1.
5.1 Transit Cost
Exhibit 5.2 below displays the transit cost knction for each of the four consolidation cases.
Each color segment represents a $20 cost range. As noted in the introduction, fi-eight logistics
exhibit returns to scale. A combination of returns to scale and reduction of fixed costs can be
clearly seen in the steep downward slopes of the cost fbnctions in the each of the cases below.
From these graphs, we can infer that longer max wait times result in larger shipment sizes and
reduced per unit transportation costs.
Exhibit 5.2: Transit Cost Graphs
ORIG
FACT
I
I
Transit Cost ( $ 1 1000kg)
FACT
OWG
Transit Cost ($ ll000kg)
Time
Max
Wait
FACT
/MaxWait Time
OMG
I
FULL
DEST
Transit Cost ($ / 1000kg)
Transit Cost ( $ 1 1000kg)
I
Max Wait Time
DEST
I1
Max Wait
Time ORlG
Max Wait
Time
M5T
I
-
'7
l
'7.
Ln
Max Wait
Time OWG
Diminishing marginal returns from consolidation are also evident in Exhibit 5.2. As max wait
times increase, the slope of the cost hnction decreases or flattens in each consolidation case. As
rnax wait times increase, more and more shipments ship because they have reached the cutoff
value rather than the rnax wait time. Opportunity for increased savings fiom consolidation
becomes saturated the longer you wait, as more and more freight reaches its consolidation
potential: a full truck or a full container.
Interestingly, under full consolidation (where port to port shipments are made primarily over one
lane) the rnax wait time at the origin port is almost inconsequential. Under full consolidation the
port to port lane is so heavy that ocean containers fill up daily. The opportunity for increased
savings fiom origin port temporal consolidation is saturated fiom day 1.
Exhibit 5.2 also shows that transit cost for destination port and full consolidation is higher across
the board when compared to factory and origin port consolidation. Exhibit 5.3 below reinforces
this finding. For each set of decision variables (DV) below, transit cost for destination port and
full consolidation is almost $100 greater than transit cost for factory or origin port consolidation.
Exhibit 5.3: Transit Cost Summary
Transit Cost ($)
DV
(1 ?I)
(1,7)
(1,141
(7,l)
(7,7)
(7,14)
(14,l)
(14,7)
(14,141
FACT
432
330
299
35 1
311
278
312
286
270
ORIG
42 1
322
297
350
301
277
310
276
26 1
DEST
512
452
436
445
405
387
42 1
395
375
FULL
498
428
405
453
399
388
45 1
405
387
The main reasons for this difference are:
a) The majority of goods in the case at hand are exported from either Shanghai or
Ningbo. These two ports are within one hour of each other. Thus, consolidating
Shanghai and Ningbo allows for increased risk pooling, but does not requires significant
cost.
b) The majority of goods in the case at hand are imported are through Chicago or Atlanta.
Consolidating shipments in LA requires a shift from short-haul inland truck to long-haul
inland truck. Consolidating at the destination port in LA is more expensive due to longhaul trucking costs.
Thus far, it appears that factory and origin port consolidation cases are superior to destination
port and full consolidation cases. Exhibit 5.4 below gives decision variable settings that result in
the minimum transit cost for each case. It should be clearly noted that the minimum cost
temporal consolidation setting for all cases is the maximum "maximum wait time" allowed in the
model, 14 days.
Exhibit 5.4: Transit Cost Savings Summary
-
-
Base Case
Minimum
% Savings
DV
-
-
-
FACT
432
270
37%
(14,14)
--
ORIG
432
26 1
39%
(24,141
DEST
432
375
13%
(14,14)
FULL
432
387
10%
(14,14)
Compared to the base case, we find potential for transit cost savings of 37% in factory
consolidation; 39% in origin port consolidation; 13% in destination port consolidation; and 10%
in h l l consolidation.
5.2 Total Transit Cost and Inventory Cost
After including inventory costs, we find that the slopes in Exhibit 5.5 are similar to those in
Exhibit 5.2. This finding indicates that inventory cost plays a minor role in the industry case. In
general, as wait times increase, transit cost should decrease and inventory cost should increase.
Y
Due to the nature of the product, however, inventory cost increases are not significant when
weighed against transit cost savings. The case would likely be different if the industry involved
fashion or high tech items that have higher value to weight ratios.
Exhibit 5.5: Transit Cost + Inventory Cost Graphs
FACT
ORIG
I
Transit Cost + Inventory Cost
Transit Cost + lnventory Cost
( $ 1 IOOOkg)
Y
s
0
Max Wait
Time
FACT
r-
Max Wait Time
OWG
ax Wait
Time
FACT
Max
I
OMG
FULL
DEST
Transit Cost + lnventory Cost
($1 IOOOkg)
Transit Cost + lnventory Cost
( $ 1 1000kg)
F
Max Wait Time
DEST
7
mT '
7.c
Max Wait
Time
OMG
I
Max Wait Time
-'-
y m ' m
7.c
m
-
Max Wait
Tim e
ORlG
Exhibit 5.6 below reinforces the finding in section 5.1, that transit costs for destination port and
full consolidation are higher across the board when compared to factory and origin port
consolidation. We again find that total transit and inventory cost for destination port and full
consolidation are approximately $100 greater than for factory or origin port consolidation.
Exhibit 5.6: Transit Cost + Inventory Cost Summary
DV
(191)
(1 ,7)
(1,141
(7,1)
(7,7)
(7,14)
(14,1)
(14,7)
(14,14)
FACT
572
473
443
507
46 1
434
475
448
440
ORIG
563
466
444
505
452
43 1
475
439
436
DEST
657
600
592
597
562
543
574
553
540
PULL
643
575
559
602
551
543
601
562
547
Transit cost and inventory cost represent the model's relevant costs. Assuming that inventory
cost inputs are correct, a decision can be made by selecting the minimum cost policy of the four
consolidation cases. Based on the minimum total inventory and transit cost found in exhibit 5.7
below, the recommended consolidation policy is origin port consolidation, factory max wait
time = 4 days, origin port max wait time = 14 days.
Exhibit 5.7: Transit Cost + Inventory Cost Savings Summary
Base Case
Minimum
% Savings
DV
FACT
572
434
24%
(7,14)
ORIG
572
429
25%
(4,14)
DEST
572
534
7%
(14,5)
PULL
572
538
6%
(9,13)
After including inventory costs, the potential for cost savings becomes more realistic, decreasing
from 37% to 24% in case FACT; 38% to 24% in case ORIG; 18% to 13% in case DEST; and
10% to 8% in case FULL. Consolidation settings that minimize transit costs are not the same
settings that minimize total transit and inventory costs!
5.3 Transit Cost, Transit Time, and Transit Time Variance (Tradeoffs)
Removing inventory cost from the equation allows us to have a crisp look at the tradeoff between
transit cost and average transit time. Exhibit 5.8 below shows a clear relationship between
transit time and transit cost: the longer the transit time, the lower the per unit transit cost.
Exhibit 5.8: Transit Cost vs. Transit Time Graph
Transit Time vs. Transit Cost
r DEST
Transit Cost 1 1000 kgs
I
1
Longer max wait times result in longer average transit times. Likewise, longer wait times result
in larger shipment sizes, which reduce the per unit transportation cost.
Exhibit 5.8 also shows that the tradeoff between transit time and transit cost continues across
spatial consolidation cases. Factory and origin port consolidation offer slower transit times at
lower costs, while destination port and full consolidation offer faster transit times at higher costs.
Most strikingly, the four cases seem to form a continuum of options along the transit time /
transit cost spectrum.
In this industry case, destination port and full consolidation hnction like different mode of
transportations. Long-haul truck service from LA expedites goods that would still be on the
water or rail, and results in dramatic improvements in transit time. As we saw in section 5.1,
transit costs are higher across the board for destination port and full consolidation. Exhibit 5.8
shows that you get what you pay for.
If air shipment was included in this study its points would extend Exhibit 5.8 to the far right,
where we would find a stream of points around the $3000 (per 1000 kg) 5 day transit time mark.
Adding sea-air, bulk air freight, and express courier transportation modes to the model would
provide a nearly continuous spectrum of transit cost / transit time options.
A key result of this model is that consolidation allows a company to fine tune the transit cost /
transit time spectrum to its liking. But the question for the company remains: What are its
priorities? How is does it value the tradeoff between cost and time? How does it pick from
among the myriad of points along the tradeoff continuum?
To attempt to answer this question, let's start by selecting the best transit time vs. transit cost
options for each of the four consolidation cases. Best options are settings that have the lowest
transit time for a given cost; they are the points make up the bottom of the transit cost / transit
time curve. Second, let's examine transit time variance for each of the settings to determine if
the best options in the transit cost / transit time tradeoff are also the best options in the transit
cost / transit time variance tradeoff.
Exhibit 5.9 below lists five or six best options for each consolidation case. We will map these
same settings against the Transit Cost 1 Transit Time Variance tradeoff in the next exhibit.
Exhibit 5.9: Transit Cost vs. Transit Time (Best Options Graph)
ORIG
FACT
-
Transit Time vs. Transit Cost
Transit Time v s Transit Cost
$260
$350
$250
$450
$300
$400
Transit Cost I 1000 kgs
Transit Cost 1 1000 kgs
--
/
$350
-
DEST
PULL
Transit Time vr Transit Cost
Transit Time vs. Transit Cost
35
h
V)
8
30
!i
i=
'g
Y
25
E?
C
20
$350
$400
$450
$500
Transit Cost 1 1000 kgs
$350
$400
$450
$500
Transit Cost / 1000 kgs
$550
In Exhibit 5.10 below we find that the settings selected as best options for the transit cost / transit
time tradeoff do not correspond as best options in the transit variance / transit time tradeoff. In
fact, in many cases the best option for the transit cost / transit time tradeoff is nearly the worst
option for the transit variance / transit time tradeoff. Considering that inventory costs increase as
transit time increases and also as transit time variance increases, we are left to balance a third
tradeoff: the transit time / transit variance tradeoff.
Exhibit 5.10: Transit Cost vs. Transit Time Variance (Best Options Graph)
FACT
ORIG
Transit Time Variance vs. Transit Costs
STDEV of Transit Time vs. Transit Costs
8
8
h
U)
h
to
h
*7
7
FI1
0
e6
E5
i=
p-6
zE
*
.-*
I-
3
*
2
*
2;
5
I-
3
w-
i*
2
V)
i!
.I-
0 1
0
$250
b
$300
$350
$400
$450
0
$250
$300
$350
$400
Transit Costs I 1000 kgs
Transit Costs 1 1000 kgs
DEST
FULL
Transit Time Variance vs. Transit Costs
Transit Time Variance vs. Transit Cost
8
h
!i
&E 6
j=
5
5
4
i
c
r
3
i
9 :,
$350
$400
$450
$500
Transit Costs 1 I000 kgs
$550
$450
$350
$400
$450
$500
Trans%Costs I 1000 kgs
So where does this leave us? First, we know that the tradeoff between transit cost and transit
time is alive and well in the world of logistics. Consolidation only allows you to fine tune the
frequency of the transit cost vs. transit time spectrum. Without a strong understanding of the
value of transit time and transit time variance, consolidation is likely irrelevant. At the same
time, if the drivers of inventory cost are known consolidation can result in significant savings
from the base case: 24% savings in the case at hand!
6. Summary of Results
>
Simulation modeling provides a clear picture of the implications of different
consolidation rules.
>
As the time allowed for consolidation increases, transit times increase and per unit freight
rates decrease.
>
The tradeoff between time and cost is seen among and across consolidation cases.
>
If the drivers of pipeline and safety stock inventory costs are known, simulation modeling
can be used to determine the best consolidation rules.
>
In the case at hand, we find 24% savings when comparing the best rules to the base case,
an immediate ship policy.
>
If the drivers of pipeline and safety stock inventory costs are unknown, or if the products
being shipped are diverse in cost and importance, simulation modeling can still provide a
clear picture of the tradeoff between transit time and transit cost.
>
In the case at hand, we find that destination port consolidation fbnctions like a different
mode of transportation. As long-haul truck service from LA expedites goods that would
still be on the water or rail, and results in dramatic improvements in transit time as well
as dramatic increases in transit cost.
The best option for a given transit time vs. transit cost tradeoff is usually not the best
option for the transit variance vs. transit cost tradeoff.
>
Without a strong understanding of the value of transit time and transit time variance,
consolidation is difficult to implement effectively.
7. Future Research
Future research interests in this area are threefold.
>
To introduce logic into the model that restricts shipments to complete orders only. The
current logic allows for split orders and thus overestimates the advantages to larger cutoff
values.
To analyze the effect of ordering policies on consolidation. It is my hypothesis that more
frequent orders to the factory will reduce cycle stock inventory costs and decrease the
variance in lead time for consolidation. Lean production with consolidation will allow
you to achieve economies of scale in transportation, while at the same time reducing the
cycle stock.
>
To add product classification logic to the model. The current logic assumes that each DC
holds products that similar in their value / weight ratio. By adding product classification
to the model, we could achieve better insights as to how consolidation rules should vary
depending on product classification.
References
Buffa, F. P., and J. R. Munn (1989). A Recursive Algorithm for Order Cycle-time that
Minimizes Logistics Cost. Journal of the Operational Research Society, 40 (4), 367-77.
Centinkaya, S., C. Y. Lee (2002). Optimal Outbound Dispatch Policies: Modeling Inventory and
Cargo Capacity. Naval Research Logistics 49,53 1-56.
CIA. World Fact Book. Viewed March, 2006.
Qttp://www.cia.gov/cia~publications/factboo
W
Dekker, R., R. Nass, and W. van Sonderen-Huisman (1997). Distribution Management by
Means of Cutoff Order Size: A Case Study. The Journal of Operatoins Research Society,
48 (1 I), 1057-1064.
Geisler, M., and W. Steger (1963). The Combination of Alternative Research Techniques in
Logistics Systems Analysis. Management Technology, 3 (I), 68-77.
Malcolm, D. (1960). Bibliography on the Use of Simulation in Management Analysis.
Operations Research, 8 (2), 169-77.
Silver, E., D. Pyke, and R. Peterson (1 998). Inventory Management and Production Planning
and Scheduling, Third Edition. New Y ork, John Wiley and Sons.
LANEID
LO 1
LO2
LO3
LO4
LO5
LO6
LO7
LO8
LO9
L10
L11
L 12
L13
L14
L15
L 16
L17
L18
L19
L20
L2 1
L22
L23
L24
L25
L26
FACTORY
F04
F12
F18
F03
F18
F20
F36
F4 1
F08
F13
F24
F25
F26
F27
F28
F35
F37
F40
F42
F02
F06
F07
F09
F11
F14
F15
ORIGIN
PORT
OP 1
OP4
OP4
OP4
OP4
OP4
OP 1
OP4
OP4
OP 1
OP4
OP4
OP4
OP 1
OP4
OP 1
OP4
OP4
OP4
OP4
OP4
OP4
OP4
OP4
OP4
OP4
DESTINATION
PORT
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DC
DC 1
DC 1
DC2
DC3
DC3
DC3
DC3
DC3
DC4
DC4
DC4
DC4
DC4
DC4
DC4
DC4
DC4
DC4
DC4
DC5
DC5
DC5
DC5
DC5
DC5
DC5
@=a
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
16000
FACTORY
CUTOFF
VALUE
Exhibit A.l: Reference Table 1, Factory to DC (100 rows)
Appendix
FACTORY
MAX
WAIT
TIME
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
ESTIMATED
ANNUAL
WEIGHT
(KGS)
228
4090
4000
363
1370
4000
2729
1124
13697
175
1111
41 1
1739
1521
648086
43880
209423
4606
361
18583
4 16305
1045522
2472
33570
69
30750
ESTIMATED
# OF
ANNUAL
ORDERS
5
15
5
5
5
5
30
65
8
1
1
2
1
2
90
4
90
1
1
2
29
70
2
1
1
2
AVERAGE
WEIGHT /
ORDER
(KGS)
46
273
800
73
274
800
91
17
1712
175
1111
206
1739
76 1
7201
10970
2327
4606
361
929 1
14355
14936
1236
33570
69
15375
WEIGHT
PER
ORDER
(STDEV)
9
55
160
15
55
160
18
3
342
35
222
41
348
152
1440
2194
465
92 1
72
1858
287 1
2987
247
6714
14
3075
2
2
1
1
2
2
2
2
2
1
2
1
2
2
.
XXX.. .
XXX.. .
XXX.. .
XXX.. .
XXX.. .
XXX.. .
XXX., .
XXX.. .
XXX.. .
.
.
.
.
XXX..
XXX..
XXX..
XXX..
XXX..
Product
Type
Factory
Name
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Origin Port
Consolidation?
m
F01
F02
F03
F04
F05
F06
F07
F08
F09
F10
F11
F12
F13
F14
Factory
OP4
OP4
OP4
OP1
OP4
OP4
OP4
OP4
OP4
OP4
OP4
OP4
OPI
OP4
Origin
Port
F010P4
F020P4
F030P4
F040Pl
F050P4
F060P4
F070P4
F080P4
F090P4
F100P4
FllOP4
F120P4
F130P1
F140P4
Factory To
Origin Port
100
100
100
100
100
100
150
100
100
100
100
150
100
100
Truck
Costs
OP4
OP4
OP4
OP 1
OP4
OP4
OP3
OP4
OP4
OP4
OP4
OP5
OP 1
OP4
Primary
Port ID
Exhibit A.2: Reference Table 2, Factory to Origin Port (100 Rows)
DLC
SHA
SZX
SHA
SHA
SHA
DLC
SHA
SHA
NGB
SHA
SHA
SHA
SHA
Primary
Port
100
100
100
100
100
100
100
100
100
100
100
100
100
100
Truck
Costs to
Primary
Port
OP4
OP4
OP4
OP 1
OP4
OP4
OP4
OP4
OP4
OP4
OP4
OP4
OP 1
OP4
Consolidation
Port ID
SHA
SHA
SHA
DLC
SHA
SHA
SHA
SHA
SHA
SHA
SHA
SHA
DLC
SHA
Consolidation
Port
100
100
100
100
100
100
150
100
100
100
100
150
100
100
Truck Costs to
Consolidation
Port
DC
ID
DC1
DC2
DC3
DC4
DC5
DC6
DC7
DC8
DC9
DP
LAX
LAX
LAX
LAX
LAX
LAX
LAX
LAX
LAX
DP
ID
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
DP4
Destination Port
Consolidation'?
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
DP -DC Lane
DP4DC 1
DP4DC2
DP4DC3
DP4DC4
DP4DC5
DP4DC6
DP4DC7
DP4DC8
DP4DC9
ROAD
MILES
20
542
22 1
4 80
260
50
50
50
200
LTL /
1000 KG
(~5000
KGS)
29
101
57
92
62
33
33
33
54
PRIMARY PORT
Consolidation
Port
LAX
LAX
LAX
LAX
LAX
LAX
LAX
LAX
LAX
TRUCK
TRUCK
MAX
CUTOFF
WEIGHT
WAIT
16000
16000
16000
16000
16000
16000
16000
16000
16000
DECISION VARIABLES
.
Primary
Port
LAX
ATL
ATL
ATL
CHI
HOU
HOU
HOU
NYC
Reference Table 4 Continued.,
DC
NAME
xxx ...
xxx ...
xxx ...
xxx ...
xxx ...
xxx.. .
xxx ...
xxx ...
xxx ...
DC - DP LINK
LTL /
1000 KG
(>5000
KGS)
23
81
45
74
50
27
27
27
43
LTL I
1000 KG
(6000
KGS)
29
285
351
385
334
260
260
260
41 1
FTL
232
806
453
73 8
496
265
265
265
430
LTL /
1000 KG
(>5000
KG S)
23
81
45
74
50
27
27
27
43
MIN
FTL
1
1
1
1
1
1
1
1
1
FTL
23 2
2,279
2,809
3,078
2,669
2,082
2,080
2,080
3,290
STDEV
MEAN
1
2
1
1
1
1
1
1
1
1
1
1
1
I
1
1
1
1
MIN
LTL
2
2
2
2
2
2
2
2
2
STDEV
1
1
1
1
1
1
1
1
1
FTL TRANSIT
1
2
1
1
1
1
1
I
1
MIN
1
1
1
1
1
1
1
1
1
MEAN
DESTINATION TRUCK COSTS
Exhibit A.4: Reference Table 4, Destination Port to DC (9 rows)
2
3
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
STDEV
2
2
2
2
2
2
2
2
2
STDEV
MEAN
2
3
2
2
2
2
2
2
2
MEAN
MIN
2
2
2
2
2
2
2
2
2
LTL TRANSIT
h
ROAD MILES
20
CONSOLIDATION PORT
Reference Table 4 Contirrroed.,,
LTL / 1000
LTL / 1000
KG ( ~ 5 0 0 0 KG (>5000
KG S)
KG S)
$
23
$
29
FTL
$ 232
I
MIN
1
MEAN
2
FTL
STDEV
1
1
MIN
2
MEAN
3
LTL
STDEV
3
DEST PORT
DP ID
DEST PORT
ATL
DP 1
DP2
CHI
HOU
DP3
LAX
DP4
NYC
DP5
X
DP6
X
DP7
X
DP8
X
DP9
BROKERAGE
FIXED FEE
$
200
$
200
$
200
$
200
$
200
$
200
$
200
$
200
$
200
FCL TIME TO AVAILABILITY
MIN
MEAN
STDEV
1
3
1
1
1
3
1
3
1
1
4
3
1
3
2
1
2
3
1
3
2
1
2
3
1
2
3
Exhibit A.5: Reference Table 5, Destination Port (9 rows)
LCL TIME TO AVAILABILITY
MIN
MEAN STDEV
2
4
3
2
4
3
4
2
3
2
4
5
2
3
4
3
2
4
2
3
4
2
4
3
2
4
3
DC 1
DC2
DC3
DC4
DC5
DC6
DC7
DC8
DC9
XXX.. .
.
.
100
100
100
100
100
100
100
100
100
Cycle
Service
Level
0.95
$ 1 kg
25
new (a)
133
k
1.645
Reference Table 6 Continued.. ..
XXX..
XXX., .
xxx..
XXX..
.
.
XXX.. .
XXX.. .
XXX..
XXX.. .
DC ID
DC
FIXED
COST /
RECEIPT
219
SS
12
11
26
2,534
5,005
5
26
56
263
E(D)
(/day)
Exhibit A.6: Reference Table 6, DC (9 rows)
SS Holding
Costs
$
984
140
120
690
6,422,530
25,049,525
24
652
3,123
69,306
E(D)' (/day)
4317
Annual
Demand
60
60
60
60
60
60
60
60
60
E(Production
Leadtime)
Pipeline
Holding Costs
$
53
24
24
24
24
24
24
24
24
24
E(Transit
Leadtime)
84
84
84
84
84
84
84
84
84
E(Tota1
Leadtime)
6
5
28
256901
1001981
1
26
125
2772
a 2 ~
3.98
3.98
3.98
3.98
3.98
3.98
3.98
3.98
3.98
a==
0.18
0.18
0.18
0.18
0.18
0.18
0.18
0.18
0.18
r